I think you are making a mistake there. The relative stability of the carbocations is this one : benzylic>allylic>tertiary>secondary>primary. The first two carbocation's stabilities is related to pi conjugation (which of course is best in the case of bezylic radicals) while the other alkyl's stability is related to overlap with neighbouring sigma orbitals. The tertiary carbocations have three neighbouring sigma orbitals with which they could overlap and so there is a greater probability of stabilization than for a secondary or primary carbocation. radicals,the same will be very stable if there is some conjugating efect to stabilize them. That is a general rule: any substituent that can conjugate with carbocations, radicals or carbanions will increase these species' stability more than an inductive effect will do, because the + or - can be dissipated over a longer system of atoms (of course the stabilization requires correcting the anomaly of numbers of electrons in the species, that is supplying electrons to carbocations and trying to take away electrons from carbanions). A radical cand be stabilized both by donating and atracting substiuents. An electron-donating substituent will surely increase the energy of the unpaired electron in the radical but will decrease the energy of the lone pair of the electron-donating substituent (2 electrons stabilized and one destabilized, that means an overall stabilization of the system). The unpaired electron, now being of greater energy will occupy a higher energy SOMO (singly occupied molecular orbital) and will be nucleophile: it will try to react with electrophilic sites(these orbitals have higgh-energy electrons so they will be more willing to donate electrons ,thus being nucleophiles instead of accepting electrons and being electrophiles). The same for electron-taking substituents: there is no lone pair involved, just the delocalization of the unpaired electron into a greater system. this lowers the energy of the electron, placing it in a lower-energy SOMO. This way it will react only with an nucleophile, because these SOMO's are more willing to accept an electron than give it away.
I hope i was clear in my explanation. I know that my way of thinking is a little different. Please comment on my post.